Machine learning in trading: theory, models, practice and algo-trading - page 1317
You are missing trading opportunities:
- Free trading apps
- Over 8,000 signals for copying
- Economic news for exploring financial markets
Registration
Log in
You agree to website policy and terms of use
If you do not have an account, please register
I can set you a task, a real task, a necessary task. But you won't solve it. Or you'll do it when everyone else has done it for you and without you. That's the problem.
And no one needs a leader who doesn't know the subject.
:))) All right, grandpa, go to sleep. Of course I don't know how to do anything.
Problems are solved jointly, but they are interpretive. This is just a word.
However, he who understands that he can get something useful out of this "something", cooks in his own porridge by himself.
But there are those who promise to pay the programmer for the program with the output or thank with something else from the super strategy, which allegedly brings him an incredible income.
Isn't that nonsense?
:))) Okay, grandpa - go to sleep. Of course I can't do anything.
Of course you can't, not even learn. Technology is lost. And the whole topic of TP is about that.
I agree. TViMS cannot completely overpower the market. That fine line beyond which it doesn't work is the presence of non-random movements in the process. How do you mathematically define them? I've already racked my brain.
So that's what I would like to see in the MO branch - coordinated teamwork on this problem. Not on the Grail, but on a specific task - the allocation of non-random periods in BP.
At least 2 people must work in the same environment, operate with the same terms.
Example: The task is: here is a sinusoid with superimposed white noise, the neural network predicts such a thing? And so on and so forth.
Shit, but there's no such thing. Shit, don't you guys get tired of trying to do this on your own? Ugh...
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
That's right, it's the scientific approach, no one builds even an apartment building all at once, I'm not talking about a spaceship, or a hadron collider, everything is broken down into tasks that consist of smaller tasks, etc. until there is basic arithmetic left in the "leaves". Before trying to defeat Forex, which is fought by guys with trillions of dollars and the best minds on the planet, one should "train on cats", in general to understand what the algorithm can cope with what it can't, etc. But here among subscribers to signals and lovers of martingale, the approach is opposite, and fanatics (probably Cossacks with CA) very aggressively and eloquently defend it, "the exchange is simple", "the article says it all", "buy the grail for $ 100", "lost - averaged! With such it is not possible and there is no motive to argue.
Regarding the collectivization and leadership, among the quants everything is very sad, even for big bucks, as soon as naklepal that works at once favorite fund (bank, propcompany ...).The trading itself is quite antisocial in its nature, and the algo-trading is even more antisocial, because they can stupidly swipe "all that they have worked hard for" on a flash drive (by bluetooth, by mail, ftp, websocket, etc.).), they say now beaten for a such (remember Alyosha), as in the 90s, and you want the goodness of your heart ...
Me too, but on short intervals. I suspect that the long-term, long time intervals,
regularities hardly exist or are easily destroyed by external factors.
PS As an example, you found a pattern lasting 24 hours, the pattern is cool!!!, entered into a trade on it. How many external and other events can happen in a day, such that from your pattern will not remain a trace? And in training, you won't learn anything about these broken patterns, only later, when you're already in the real game. At least because of the multitude of variants of events.
The future is not defined, the patterns appear and disappear, that's normal, but the fact that they must necessarily be short-lived is questionable. I do not have a very large sample due to the trend strategy and therefore I think it is unreasonable to reduce it even further.
However, I decided to conduct an experiment on the effectiveness of training on different proportions of the training and test sample involved in the training. Step will be 10%, ie a training in the beginning of 90% and 10% of the test, then the test gradually increased by 10%, in each case will be 200 models - see what happens. Another question, how best to compare these combinations, the average or the absolute criterion - ideas are accepted.
But the trouble is, everyone here is like that - some even starve to death and sleep in a garbage can, but they don't do anything together, they don't apprentice to anyone. This is the paranormality inherent in this forum. If this fact is commonplace to you, it just strikes me to the core.
And who's sleeping by the trash can here - aren't you delusional?
The most valuable resource is time, so giving your time to an enthusiast is not a very wise thing to do. I have previously suggested an association along ideology to jointly develop ideas with which the majority of this community agrees, but that option was not viable either.
That's why I invested my time into development and implementation of my ideas - I work 12-14 hours every day, including weekends, it would be impossible to do it all, working at other jobs.
And who sleeps by the garbage can, are you not delusional?
Don't take it personally, Alexei. Do you know how many teachers there were here before you? People gave 10-15 years to the Ministry of Education, but in the end - no result, disaster, poverty... I'm not kidding - it was like that. And what is remarkable is that these people chose their own model and have been working on it for years, not listening to anyone, but on the contrary, trying to sell it to everyone, despite its obvious ineffectiveness. I remember these stories well and I personally would not like to repeat them.
If you still have strength, be my guest.
Don't take it personally, Alexei. You know how many Teachers there were before you? People gave 10-15 years to the Ministry of Defense, and in the end - no result, disaster, poverty... I'm not kidding - it was like that. And what is remarkable is that these people chose their own model and have been working on it for years, not listening to anyone, but on the contrary, trying to sell it to everyone, despite its obvious ineffectiveness. I remember these stories well and personally would not like to repeat them.
If you still have strength, be my guest.
Let me tell you, I'm teaching someone here how to do something, what does the term "Teacher" have to do with it? MO is as much an option as not MO - anyway, people then spin the idea in the optimizer and adjust to the story, and the result is deplorable. No one promised that you can live off the market. In my life I had ups and downs, everything happens cyclically enough, the only trouble is that each new ups take more energy - life limits our development in time.
Here is the problem from Doc again
Forum on trading, automated trading systems and trading strategies testing
From Theory to Practice
Dr. Trader, 2018.05.08 12:02
There are two test files in the atacha archive. Both contain values in the normal distribution, the histograms are the same and almost symmetrical with respect to zero.
But these files have one very big difference - their markness.
One file has a memory (non-markovian process), you can try to predict "the next value is greater than or less than zero" relying on past values. You can apply neuronics and other machine learning to predict.
The other file has no memory (Markov process), any prediction will fail. Machine learning is powerless, but maybe Alexander will be able to predict something with physics.
Who will learn to identify which file has memory and which does not - well done, and applying the same method to forex will finally prove that the pricing process is indeed Markovian.
It is also worth checking if the normal distribution is a sufficient condition for profitability of the model. Make a cumulative cum() random walk graph and try to trade on it.
Here he had a model that did the job and made a profit on one of the artificial series attached in that post. If your model can't do that - maybe it's too early to take on real BPs?
He was working with first differences - increments on thinned real BPs, and something else (he didn't tell me everything, of course). His model predicted the sign of the next increment and his signal was positive.
So why does everyone here spit on the experience of successful traders and reinvent everything here, scaring the youth with some herbarium! I do not understand. I refuse to understand.